Table 3 Damage identification performance comparison of different algorithms.
Models | Accuracy/% | Recall/% | Precision/% | IOU/% | FPS |
|---|---|---|---|---|---|
Proposed method | 90.05 | 88.94 | 89.22 | 88.67 | 57.74 |
Attention U-Net | 87.23 | 85.67 | 86.45 | 82.34 | 20.23 |
RefineNet | 86.45 | 84.23 | 85.34 | 81.45 | 18.12 |
HRNet | 88.12 | 87.54 | 88.34 | 84.12 | 22.34 |
DeepLabv3 | 85.67 | 83.89 | 84.56 | 80.89 | 19.89 |
SegNet | 82.34 | 80.12 | 81.23 | 77.34 | 18.67 |